Akamai signs $11.6 billion cloud computing deal with Anthropic
Read more
YourStory [india, en]
yourstory.com

Akamai signs $11.6 billion cloud computing deal with Anthropic

Akamai Technologies has signed a seven-year agreement worth $11.6 billion with the artificial intelligence developer Anthropic. This agreement could potentially be increased by another $9 billion, bringing the total to $20 billion.

This partnership signals the growing demand for powerful computing resources due to the global scaling of AI tools. Instead of focusing on Graphics Processing Units (GPUs), which currently drive the AI industry, the collaboration is centered on Central Processing Units (CPUs).

Thanks to this deal, Anthropic will be able to utilize Akamai's distributed AI infrastructure and software to handle growing demands for CPU computing power and provide its AI applications more quickly and reliably.

This transaction is a significant success for Akamai as the company accelerates its expansion beyond traditional content delivery and cybersecurity into large-scale cloud computing. Prior to this announcement, Akamai had recorded only slightly over $2.8 billion in multi-year commitments for cloud infrastructure at the beginning of the current year.

Dr. Tom Leighton, co-founder and CEO of Akamai, stated: 'Anthropic is driving the AI revolution, and we are pleased they have chosen Akamai's capabilities to build and operate AI infrastructure at scale.' He added that 'our expanding global network combined with years of experience serving the world's largest enterprises allows us to act as an infrastructure provider for secure and responsible AI applications.'

Under the partnership, Akamai grants Anthropic an equity stake in its results. Akamai issued Anthropic a warrant to purchase 7.7 million non-voting Series B preferred shares convertible into common stock, or approximately 5% of Akamai's total outstanding shares, at an exercise price of $111.33 per share.

Approximately 2% of these shares will be held under the initial $11.6 billion agreement. The remaining 3% will be held incrementally—about 1% for every additional $3 billion spent on cloud services if Anthropic decides to expand the contract over the next seven years.

The total capital expenditure required to support such a large volume of computing is estimated at around $5.5 billion. In a statement, Akamai indicated that it does not expect an impact on its 2026 revenue forecast but projects an increase in capital expenditures of about $1.7 billion in 2026 to ensure and pre-purchase critical supply chain components, including memory.

Akamai's cloud platform provides high-performance cloud computing from core to edge, enabling organizations to build and scale next-generation AI applications while providing comprehensive multi-layered protection to defend enterprises against evolving cyber threats.

Similar stories

Alibaba Cloud introduces agent cloud solutions stacks: AgentCore, next-generation CPFS, and HPN 8.0 Pro
Read more
pandaily.com

Alibaba Cloud introduces agent cloud solutions stacks: AgentCore, next-generation CPFS, and HPN 8.0 Pro

At the Apsara Cloud 2026 conference, Alibaba Cloud presented its agent cloud strategy, which is based on three scenarios: model, management, and context. These scenarios correspond to three levels: AI Native Cloud, Agent Native Cloud, and Context Engine, according to the company's press release from September 22.

This review focuses primarily on the products of the agent platform and cloud infrastructure, rather than the roadmaps for training base models, which are covered in other materials.

At the management level, Alibaba Cloud announced AgentCore—an enterprise platform designed for creating, launching, and administering AI agents throughout their entire lifecycle. According to the company's materials, this platform allows organizations to gather tools, control human-agent interaction, track performance, and implement security at the operational level. This ensures the integration of diverse agents into critical business systems under a single controlled foundation. Additionally, the Agent Security Center was introduced, which ensures security and compliance throughout the entire lifecycle with real-time threat detection.

Chinese trade media commentary on the technical director of Alibaba Cloud indicated the vendor's claims that AgentCore can improve task completion rates and reduce total cost of ownership, although these figures remain company claims.

Regarding the model serving subsystem, Alibaba Cloud announced the launch of a next-generation Parallel File System (CPFS). This system provides throughput of hundreds of terabytes per second and one hundred million input/output operations per second, which, according to the company, can reduce enterprise AI storage costs by approximately 69 percent. In terms of internal storage, it was noted that the scale of one file system reached about 100 Petabytes, five times the previous figure. Furthermore, positive effects for the training process were noted, such as faster job startup and higher utilization of peak accelerators.

The HPN 8.0 Pro network architecture was separately presented, representing Alibaba Cloud's latest patented AI architecture. It is capable of providing 100 petabits of throughput with ultra-low latency. One cluster can support over 130,000 network ports at 800G speed, while redundancy ensures that failures in optical transceivers or communication lines do not interrupt service. The impact of updates and network failures has been reduced from 50 percent to 25 percent compared to the previous generation.

Collectively, the combination of AgentCore, next-generation CPFS, and HPN 8.0 Pro forms a specific product triad of cloud solutions that Alibaba Cloud demonstrated for scalable agent workloads, independent of chip announcements or large language models made during the same week of the conference.

Popular